If you want to know how to improve inventory accuracy, start by measuring it properly, then work backwards from the errors to the steps that cause them. Stock records drift from the shelf in a handful of places: receiving, putaway, picking, returns and unit-of-measure set-up. Most of the gain comes from making each movement a recorded event rather than a note on a clipboard, and from counting small samples often instead of relying on one big stock take a year.
The stakes are real for UK operators. The UK Warehouse Association speaks for the sector’s warehouse businesses, and food operators face traceability expectations under UK food law and retailer standards such as BRCGS. Both depend on stock records you can trust.
What inventory accuracy measures and why it matters
Inventory accuracy is the share of stock records that match what is physically in the warehouse when you count. A location holding the right quantity counts as accurate, and so does an item line. The two measures answer different questions, so choose the one that fits the problem you are trying to solve.
Terminology varies between businesses. The ASCM publishes supply chain terms that make a useful starting point, but you still need your own definition. Write it down before you count, with three things: what counts as a match, how much variance you tolerate, and whether a check covers every location or a sample. Logistics UK covers the wider sector, but inventory accuracy is an internal measure. It tells you whether the number on the screen is safe to promise to a customer.
Location accuracy and item accuracy
Location accuracy asks whether each bin, rack or pallet position holds the quantity the system says it does. It finds misplaced stock and putaway errors, and it matters most where stock moves between many locations.
Item accuracy asks whether each SKU matches the system quantity across the whole site. It finds short and over-stocked lines, and it is the measure that tells you whether you can promise a product to a customer.
How to calculate inventory accuracy
Use the same working for both measures, then multiply by 100 to get a percentage.
Location accuracy = locations counted that match the system ÷ total locations counted × 100
Item accuracy = SKU lines counted that match the system ÷ total SKU lines counted × 100
Take an example count of 480 locations, where 463 match the system. That gives 463 ÷ 480 × 100, or 96.5% location accuracy. If you count 2,140 SKU lines and 2,071 match exactly, item accuracy is 96.8%. These example figures are invented to show the working, not data from any warehouse.
Two habits keep the number honest. Count the same way each time, with the same tolerances and the same cut-off point. And record the reason for every variance, because one percentage hides whether errors are random or come from a single broken step.
Where stock records drift from the shelf
Stock errors rarely appear where they start. A short pick is often a putaway or receiving mistake that nobody spotted for a week. The table below links common symptoms to the step where the error usually begins.
| Symptom | Likely cause | Where to look |
|---|---|---|
| System shows stock that cannot be found | Putaway without a location scan, or a move never recorded | Plassering |
| Short picks on lines that should have stock | Wrong location or pallet confirmed, or quantity keyed by hand | Plukking |
| Stock on the shelf that the system never received | Putaway before receipt is finished, or quantity typed from paperwork | Receiving |
| Phantom stock after returns | Returns restocked without a booking or inspection | Returns |
| Variance that looks like random loss | Cases counted in one unit and held in another | Unit of measure |
Receiving
Receiving is where a wrong number first enters the system. Mathew Buttar, Head of Solutions at Clarus, says unannounced stock arrives all the time. When the dock backs up, the quickest route is to take the quantity from the paperwork rather than count the pallet, and then the system holds stock that nobody checked in.
Scan the pallet instead of keying it. Where a supplier sends an advance shipping notice with the right identifiers, the pallet label can carry the product, batch and quantity in one scan. Clarus’s handheld goods-in flows can read standard GS1 labels this way, so the receiver confirms what is on the pallet rather than retyping it. Record damage at the dock too. A broken item is far easier to write off on the day it arrives than to find in a stock take months later.
Plassering
Putaway errors look like missing stock. The pallet went somewhere sensible, but nobody scanned the location, so the system still shows it in the receiving bay. Scan the location before you confirm the putaway, so the record and the rack agree before the forklift moves on.
Bulk storage creates its own problem. First-in, first-out works well in racking, but in deep bulk stacks the oldest stock sits at the back. Mathew Buttar’s rule of thumb is that dated product should stay accessible pallet by pallet, because otherwise staff end up shuffling pallets to reach the oldest one. Damaged pallets also need restacking before they go back into racking, and the Health and Safety Executive is the place to check your storage and racking routine.
Plukking
Picking is where the visible errors show up, because a short pick is the one the customer notices. The usual causes are a picker taking from the wrong location, a quantity keyed by hand, and a paper pick list nobody double-checks.
Scanning the location, the pallet and the product before each pick is confirmed catches most of these at the shelf. If someone scans the wrong location, the handheld should stop the task and show where the right one is. Mathew Buttar is clear about the limit, though. A scan cannot see a picker who lifts a box from the pallet beside the one they scanned, so training and spot checks still matter.
Paper pick lists are the weakest link. In a paper operation, one person picks, a second person checks the pallet against the list, and then the order is wrapped and dispatched. That is two people’s labour to verify one pick. Scanned picking does the check at the point of pick, so the second pair of eyes is no longer the only control.
Returns
Returns are a common source of phantom stock. Goods come back, go onto a shelf and get sold again before anyone has inspected them, or the return is never booked in at all. Treat returns as a receipt, not as an exception. Book every item in, hold it in an inspection location, and release it to sellable stock only once someone has checked it.
Unit of measure
Unit-of-measure errors are quiet and expensive. Stock is received by the case, stored by the pallet and picked by the each, and somewhere between those steps a conversion goes wrong. The variance then looks like random loss, when it is really one pack size being counted as another. Set one base unit for each product, define every pack as a multiple of it, and test the conversions with real stock before you rely on them.
Fixes by area, in the order errors happen
Work through the process from the dock to the despatch door, and fix the step that creates errors before you tune the step that catches them.
- Mottak: Scan pallets rather than keying them, match against the advance notice, and record damage at the dock.
- Putaway: Scan the location before confirming, and keep dated stock accessible.
- Plukking: Scan the location, pallet and product, and move away from paper lists where you can.
- Returns: Book every return in, inspect it, then release it.
- Unit of measure: Use one base unit per product and test every pack conversion.
Count often enough to catch drift early
Cycle counting catches drift while the cause is still fresh, and it avoids the disruption of a full stock take. A full count means stopping operations, paying for overtime or running night shifts. Mathew Buttar’s preferred method is a rolling sample. Count a set number of locations each week, measure the error rate on those, and use it to judge the rest of the stock until the next count. In his words, “Cycle counting is kind of where you want to be.”
Count the fast-moving locations more often, and count any location again after a known error. Set your own accuracy target from what an error costs you, because no single figure suits every operation. If a customer or an audit sets one, that figure wins. If you are planning a full stock take for the year end, plan it as a separate event rather than the only check you run.
Clarus customers show what a sustained routine can do. JODA Freight rose from the 90% to 97% range in year one to 99% stock accuracy by year two, and its stock takes moved from weeks of Excel work to one day, with year-end counts in two to three days, according to its customer story. RTS reports 99% stock accuracy across its European operation on cloud WMS, as set out in its customer story. These are customer results, not industry benchmarks.
Tools that help: scanning, real-time records and counting
Tools only help when the process behind them is sound, but three types of tooling do most of the work.
- Barcode scanning at every movement. Scanning turns the location, pallet and product checks into part of the task rather than an extra step. Handheld workflows can be set up per process, so receiving, putaway and picking each ask for what they need.
- Real-time stock records. A task-based system records each movement as it is completed. Clarus works at line level, so every pick and receipt task updates the record as it happens, and the inventory management features keep those records in one place.
- Structured stock checks. Cycle counts run as set tasks rather than ad hoc walks. The guide to running structured stock checks covers how to plan them around live operations.
Track the result alongside the rest of the operation. The warehouse KPIs worth reviewing each week include accuracy alongside picking and despatch measures, and the good warehouse practices checklist gives your team a floor-level list to work through.
Snakk med en lageransvarlig
Hvis du vurderer ulike alternativer og ønsker å se hvordan et spesialutviklet lagerstyringssystem (WMS) fungerer i praksis, er Clarus verdt å ta en prat med. Vi samarbeider med 3PL-leverandører og distributører over hele Storbritannia for å implementere lagerstyringsprogramvare som tilpasses din driftsmodell – ikke omvendt.
Ta kontakt med teamet vårt å snakke gjennom dine krav.
Related reading: how cycle counting works og real time inventory tracking.